## Box 1. Network Analysis: Selection Menu in Program Code

## Source details

**Canonical URL:** [Box 1. Network Analysis: Selection Menu in Program Code](https://www.imf.org/-/media/files/publications/wp/2017/wp17216.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2017/wp17216.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2017/wp17216.pdf.json)

---

### Purpose and software environment
- Analytical tool is programmed in the R programming language.
- Uses the open source package "igraph" for network analysis.
- R chosen because it is open source, supports network analysis via "igraph", and can hold, process, and analyze large amounts of data (e.g., Excel can hold only about 1 million rows).
- Only the consolidated SWIFT data file in CSV format is needed to run the analytical tool.

### Data format and pre-processing requirements
- Input must be a consolidated SWIFT CSV file (transaction-level).
- Use CSV rather than XLS to: (i) save space (CSV stores only cell values); and (ii) ensure readability in many programs including R.
- Reporting banks must follow the data format requirements exactly; deviations (even trailing spaces) will be treated as different identifiers by the code (example: "BOFAUS33" vs "BOFAUS33 ").
- Global correspondent banks’ BIC8s must be clustered (Annex I.C); this can be done with Excel’s Find and Replace (CTRL+F on Windows).
- If SWIFT data are collected from different reporting banks, the SWIFT data must be consolidated into one CSV file (one tab); when saving as CSV from Excel, formulas and formatting are lost and only the first tab is saved.

### Menu selection options in the program code (lines and selectable values)
- Menu is implemented in the provided R code; key selection lines and selectable values are:
  - Line 110: select which part of the payment chain to analyze:
    - "f" = full payment chain (initial ordering—correspondent bank—ultimate beneficiary bank)
    - "e" = beginning/end point analysis (reporting bank—initial ordering/ultimate beneficiary bank; excludes correspondent bank)
    - "r" = flows between reporting bank and correspondent bank only
  - Line 111: select analysis scope:
    - "s" = full system analysis (including all reporting banks)
    - "b" = individual bank level analysis (requires BIC8 input)
  - Line 112: select aggregation level:
    - "c" = country level (countries in which ordering, beneficiary and correspondent banks are domiciled)
    - "i" = institution level (actual initial ordering, ultimate beneficiary and correspondent bank)
  - Line 113: when country-level ("c") selected, index-based coloring option:
    - "yes" = enable index-based coloring of different countries
  - Line 114: when "b" (bank-level) selected, provide the BIC8 code of the reporting bank to be analyzed.
  - Line 116: select the period to analyze (example given: period 6 corresponds with 2016).
  - Line 117: select the currency of the SWIFT transactions to be analyzed.

### Code selection options (explicit code variables)
- Line 110 of the code: pathway_length_toggle <- "f" # Can be r for "correspondent relationships" e for "endpoint" (ordering and end beneficiary) or f for "full"
- Line 111 of the code: system_or_individual_view <- "s" # Can be s for "system" or b for "bank"
- Line 112 of the code: country_or_institution_view <- "c" # Can be c for "country" or i for "institution"
- Line 113 of the code: color_using_index <- “yes” # Can be “yes” or “no”. Requires an additional file if “yes”.
- Line 114 of the code: selected_bank <- "XXXYYYZZ"            # Needs to be a BIC 8-letter code
- Line 115 of the code: selected_country <- "XU"                  # Needs to be an ISO 2-letter code for the country of the respondent bank(s) under analysis
- Line 116 of the code: selected_period <- “6” # Selects the period (1-6) for which the map will be generated
- Line 117 of the code: currency <- "USD"                         # Needs to be a 3-letter currency ticker, use "All" if not specifying a single currency

### Network analysis overview and centrality measures
- Using igraph, network analysis is performed on SWIFT data to provide qualitative (visual) and quantitative analysis of relationships between financial institutions.
- Financial institutions are represented as nodes and flows as links; node size, shape, and color and edge size and color can denote multiple types of information.
- All nodes correspond to a BIC8 (which can be made visible) when institution-level displays are used.
- Centrality measures described:
  - Betweenness: A node with high betweennness lies on a high proportion of paths
  - Closeness: A node with high closeness can transact quickly with nodes in the network.
  - Degree: A node with high degree has many relationships.
  - Eigenvector: A node with a high eigenvector is connected to important neighbors.
  - Note: PageRank is a simplified version of Eigenvector centrality, taking less time to compute (Brin; Page, 1998).

### Practical insights, detection capabilities, and visualization encodings
- New flows (compared with the previous reporting period) between correspondent banks and initial ordering or beneficiary banks are colored green.
- Comparing network maps of consecutive reporting periods allows detection of:
  - whether the network has grown or shrunk;
  - changes in the relative importance of major correspondent banks;
  - new correspondent relationships.
- Visualization encodings (examples):
  - Node size reflects PageRank score.
  - Edge width reflects normalized flow size.
  - Orange nodes reflect domestic institutions (in examples).
  - Green edges reflect new relationships compared to previous period.
  - Node labels (BIC8s or ISO 2-letter country codes) may be turned off for visibility.

### Examples and exact selection combinations for figure generation
- Figure 6 (2016 Institution-Level, System (All U.S. Dollar Relationships)): select “f,” “b,” “i,” and “USD”.
- Figure 7 (2016 System-Level U.S. Dollar Correspondent Relationships): select “r,” “s,” “i,” and “USD.”
- Figure 8 (2016 System-Level Euro Correspondent Banking Relationships): select “r,” “s,” “i,” and “EUR.”
- Figure 9 (2016 System-Level All Correspondent Banking Relationships): select “r,” “s,” “i,” and “All.”
- Figure 10 (2016 Bank-Level Correspondent Relationships): select “r,” “b,” "i,” and “USD.”
- Figure 11 (2016 Country-Level, System (All Relationships)): select “f,” “s,” “c,” the country code of the respondent bank, “USD,” and “no” for using an additional index.
- Figure 12 (2016 Country-Level, System (All Relationships) with index): select “f,” “s,” “c,” the country code of the respondent bank, “USD,” and “yes” for using an additional index (e.g., Corruption Perception Index in Annex IV).

### Use of additional index and customization
- color_using_index can be “yes” or “no”; when “yes” an additional CSV file with country-based data (e.g., CPI) is required.
- With an additional index, countries receive colors and rim colors based on index quintile and change from previous period:
  - Inside node color: white for domestic country, orange and red for second-lowest and lowest quintiles, respectively (position in the relevant period’s CPI).
  - Outside node color: green if the country node’s CPI score improved, red if it worsened over the previous period.
  - Additional CPI-based encoding: red (countries in the lowest quintile on the CPI), orange (countries in the 3rd quintile of the CPI), grey (all other countries). Outer ring colored green (improved), black (same ranking), or red (lower ranking) compared to previous year.
- The underlying code can be modified to:
  - incorporate different country-based information (examples listed: Macroeconomic data, various country risk indicators, cyber risk threat data);
  - simulate the impact of various scenarios (similar to a stress test), though scenario simulation is beyond the scope of the paper.

### Practical notes, tradeoffs, and caveats
- Gathering SWIFT data directly from SWIFT incurs a fee; gathering from institutions requires strict adherence to template naming conventions (BIC8, ISO two-letter currency abbreviations).
- The analytical tool can generate all basic network maps from the consolidated SWIFT CSV file.
- Advanced network maps can be produced by combining SWIFT data with other index- and country-based data (e.g., Corruption Perception Index), but this requires code modifications and programming capacity.
- Changes to the template format in Annex I.B will affect the functioning of the R code; the code expects the template format as provided.
- The Expanded Scope data can be used to cross-check Minimum Scope Template results but requires additional collection cost and capacity.
- Annex I.A and Annex I.B provide Minimum and Expanded Scope Template formatting guidance (including formatting constraints such as numeric fields not including commas/decimals and date fields text-formatted like Mar-17).

---

### ANNEX II: PROGRAMMING CODE ANALYTICAL TOOL FOR SWIFT DATA ANALYSIS

### Network analysis: qualitative and quantitative mapping of financial institutions
- Network analysis represents financial institutions as nodes and flows between them as edges; node size is used to represent a node’s centrality score.
- Centrality indicates critical positions in the network and different centrality measures capture different types of importance.
- Practical insights from the Krackhardt kite graph example (ten nodes, banks numbered 0–9, each edge reflects the same amount):
  - Bank 3 has the highest degree (most relationships).
  - Failure of Bank 3 would not disconnect payments between other banks.
  - Bank 7 has the highest betweenness; its failure would break flows between Banks 8 (and 9) and the rest of the network.
  - Banks 5 and 6 have the highest closeness; flows from them take the fewest steps to reach all points in the network.
- Implementation note: While the Extended Scope Framework uses PageRank as a metric for the relative size of nodes, any centrality measure could be used instead depending on which information about the banks is most useful.

### Country-level overlay data: CPI list used for country data analysis
- Purpose: Add country-based information (for example, Corruption Perceptions Index (CPI) scores) as layers onto country network maps to display multiple pieces of information simultaneously (e.g., node interior color from yearly CPI scores and rim color from yearly changes).
- Data structure described:
  - A .csv file containing the ranking of 176 countries on the Corruption Perceptions Index (CPI) published by Transparency International.
  - First column: ISO two-letter country code (to be merged with SWIFT country data).
  - Columns s2012 – s2016: yearly ranking of the countries (0 when no score is available).
  - Columns ch2013 – ch2016: change in the score compared with the previous year (0 if same, 1 if score increased, -1 if the score decreased, and empty cell if no score was available).
- Sample CPI table header and rows as presented in the source (preserving formatting and values):
  - Codes2012s2013s2014s2015s2016ch2012ch2013ch2014ch2015ch2016
  - AF88121115001-11
  - AL33313336390-1111
  - DZ34363636340100-1
  - AO222319151801-1-11
  - AR35343432360-10-11
  - AM3436373533011-1-1
  - AU85818079790-1-1-10
  - AT69697276750011-1
  - AZ272829293001101
  - BS717171066000
  - BH51484951430-111-1
  - BD262725252601-101
  - BB7675740610-1-1
  - BY31293132400-1111
  - BE757576777700110
  - BJ3636393736001-1-1
  - BT636365656500100
  - BO3434353433001-1-1
  - BA424239383900-1-11
  - BW65646363600-1-10-1
  - BR43424338400-11-11
  - BN5560005801
  - BG4141434141001-10
  - BF383838384200001
  - BI192120212001-11-1
  - KH22202121210-1100
  - CM26252727260-110-1
  - CA84818183820-101-1
  - CV60585755590-1-1-11
  - CF26252424200-1-10-1
  - TD19192222200010-1
  - CL72717370660-11-1-1
  - CN394036374001-111
  - CO363637373700100
  - KM282826262400-10-1
  - CG26222323200-110-1
  - CR54535455580-1111
  - CI29273232340-1101
  - HR46484851490101-1
  - CU48464647470-1010
  - CY66636361550-10-1-1
  - CZ49485156550-111-1
  - CD21222222210100-1
  - DK9091929190011-1-1
  - DJ363634343000-10-1
  - DM585858059000

*Source: wp17216 - Box 1 and ANNEX II: Programming Code Analytical Tool for SWIFT Data Analysis*

### References .............................................................................................................

### References

### Annexes
- Annex I.A: Minimum Scope Template ..................................................................................... 38
- Annex I.B: Expanded Scope Template ..................................................................................... 40
- Annex I.C: Clustering of Business Identifier Codes ................................................................. 41
- Annex II: Programming Code Analytical Tool for Swift Data Analysis.................................. 42
- Annex III: Background Network Analysis ............................................................................ 47
- Annex IV: CPI List Used for Country Data Analysis ........................................................... 48

### Figures
- Figure 1: Stylized Cross-Border Payment Chain....................................................13
- Figure 2: Stylized Nested Cross-Border Payment Chains ..............................................14
- Figure 3: System-Level AC and CBR Development...............................................20
- Figure 4: Bank-Level (B3) AC and CBR Development...........................................23
- Figure 5: Analyzing Remittances Flows.............................................................26
- Figure 6. 2016 Institution-Level, System (All U.S. Dollar Relationships)......................30
- Figure 7. 2016 System-Level U.S. Dollar Correspondent Relationships........................31
- Figure 8. 2016 System-Level Euro Correspondent Banking Relationships.....................32
- Figure 9. 2016 System-Level All Correspondent Banking Relationships........................32
- Figure 10. 2016 Bank-Level Correspondent Relationships.........................................33
- Figure 11. 2016 Country-Level, System (All Relationships) ......................................34
- Figure 12. 2016 Country-Level, System (All Relationships)......................................35

### Boxes
- (No boxes listed)

*Source: wp17216 - References (PDF).*

### Box 1. Network Analysis: Selection Menu in Program Code.....................................29

### Box 1. Network Analysis: Selection Menu in Program Code

### Purpose and software environment
- Analytical tool is programmed in the R programming language.
- Uses the open source package "igraph" for network analysis.
- R chosen because it is open source, supports network analysis via "igraph", and can hold, process, and analyze large amounts of data (e.g., Excel can hold only about 1 million rows).
- Only the consolidated SWIFT data file in CSV format is needed to run the analytical tool.

### Data format and pre-processing requirements
- Input must be a consolidated SWIFT CSV file (transaction-level).
- Use CSV rather than XLS to: (i) save space (CSV stores only cell values); and (ii) ensure readability in many programs including R.
- Reporting banks must follow the data format requirements exactly; deviations (even trailing spaces) will be treated as different identifiers by the code (example: "BOFAUS33" vs "BOFAUS33 ").
- Global correspondent banks’ BIC8s must be clustered (Annex I.C); this can be done with Excel’s Find and Replace (CTRL+F on Windows).

### Menu selection options in the program code (lines and selectable values)
- Menu is implemented in the provided R code; key selection lines and selectable values are:
  - Line 110: select which part of the payment chain to analyze:
    - "f" = full payment chain (initial ordering—correspondent bank—ultimate beneficiary bank)
    - "e" = beginning/end point analysis (reporting bank—initial ordering/ultimate beneficiary bank; excludes correspondent bank)
    - "r" = flows between reporting bank and correspondent bank only
  - Line 111: select analysis scope:
    - "s" = full system analysis (including all reporting banks)
    - "b" = individual bank level analysis (requires BIC8 input)
  - Line 112: select aggregation level:
    - "c" = country level (countries in which ordering, beneficiary and correspondent banks are domiciled)
    - "i" = institution level (actual initial ordering, ultimate beneficiary and correspondent bank)
  - Line 113: when country-level ("c") selected, index-based coloring option:
    - "yes" = enable index-based coloring of different countries
  - Line 114: when "b" (bank-level) selected, provide the BIC8 code of the reporting bank to be analyzed.
  - Line 116: select the period to analyze (example given: period 6 corresponds with 2016).
  - Line 117: select the currency of the SWIFT transactions to be analyzed.

### Practical notes, tradeoffs, and caveats
- Gathering SWIFT data directly from SWIFT incurs a fee; gathering from institutions requires strict adherence to template naming conventions (BIC8, ISO two-letter currency abbreviations).
- The analytical tool can generate all basic network maps from the consolidated SWIFT CSV file.
- Advanced network maps can be produced by combining SWIFT data with other index- and country-based data (e.g., Corruption Perception Index), but this requires code modifications and programming capacity.
- Changes to the template format in Annex I.B will affect the functioning of the R code; the code expects the template format as provided.
- If SWIFT data are collected from different reporting banks, the SWIFT data must be consolidated into one CSV file (one tab); when saving as CSV from Excel, formulas and formatting are lost and only the first tab is saved.

*Source: wp17216 - Box 1. Network Analysis: Selection Menu in Program Code*

### Box 1. Network Analysis: Selection Menu in Program Code

### Box 1. Network Analysis: Selection Menu in Program Code

### Code selection options
- Line 110 of the code: pathway_length_toggle <- "f" # Can be r for "correspondent relationships" e for "endpoint" (ordering and end beneficiary) or f for "full"
- Line 111 of the code: system_or_individual_view <- "s" # Can be s for "system" or b for "bank"
- Line 112 of the code: country_or_institution_view <- "c" # Can be c for "country" or i for "institution"
- Line 113 of the code: color_using_index <- “yes” # Can be “yes” or “no”. Requires an additional file if “yes”.
- Line 114 of the code: selected_bank <- "XXXYYYZZ"            # Needs to be a BIC 8-letter code
- Line 115 of the code: selected_country <- "XU"                  # Needs to be an ISO 2-letter code for the country of the respondent bank(s) under analysis
- Line 116 of the code: selected_period <- “6” # Selects the period (1-6) for which the map will be generated
- Line 117 of the code: currency <- "USD"                         # Needs to be a 3-letter currency ticker, use "All" if not specifying a single currency

### Network analysis overview
- Using igraph, network analysis is performed on SWIFT data to provide qualitative (visual) and quantitative analysis of relationships between financial institutions.
- Financial institutions are represented as nodes and flows as links; node size, shape, and color and edge size and color can denote multiple types of information.
- All nodes correspond to a BIC8 (which can be made visible) when institution-level displays are used.
- New flows (compared with the previous reporting period) between correspondent banks and initial ordering or beneficiary banks are colored green.
- Comparing network maps of consecutive reporting periods allows detection of:
  - whether the network has grown or shrunk;
  - changes in the relative importance of major correspondent banks;
  - new correspondent relationships.

### Examples and figure generation (exact selection combinations)
- Figure 6 (2016 Institution-Level, System (All U.S. Dollar Relationships)) is generated by selecting “f,” “b,” “i,” and “USD” in the menu in R.
  - Visualization notes from Figure 6: Node size reflects PageRank score (Brin; Page, 1998), edge width reflects normalized flow size; orange nodes reflect domestic institutions, green edges reflect new relationships compared to previous period. Node labels (BIC8s) have been turned off for visibility.
- Figure 7 (2016 System-Level U.S. Dollar Correspondent Relationships) is generated by selecting “r,” “s,” “i,” and “USD.”
  - Visualization notes from Figure 7: Node size reflects PageRank score, edge width reflects normalized flow size; orange nodes reflect domestic institutions, green edges reflect new relationships compared to previous period. Node labels (BIC8s) have been turned off for visibility.
- Figure 8 (2016 System-Level Euro Correspondent Banking Relationships) is generated by selecting “r,” “s,” “i,” and “EUR.”
  - Visualization notes from Figure 8: Node size reflects PageRank score, edge width reflects normalized flow size; orange nodes reflect domestic institutions, green edges reflect new relationships compared to previous period. Node labels (BIC8s) have been turned off for visibility.
- Figure 9 (2016 System-Level All Correspondent Banking Relationships) is generated by selecting “r,” “s,” “i,” and “All.”
  - Visualization notes from Figure 9: Node size reflects PageRank score, edge width reflects normalized flow size; orange nodes reflect domestic institutions, green edges reflect new relationships compared to previous period. Node labels (BIC8s) have been turned off for visibility.
- Figure 10 (2016 Bank-Level Correspondent Relationships) is generated by selecting “r,” “b,” ‘i,” and “USD.”
  - Visualization notes from Figure 10: Node size reflects PageRank score, edge width reflects normalized flow size; orange nodes reflect domestic institutions, green edges (in this case none) reflect new relationships compared to previous period. Node labels (BIC8s) have been turned off for visibility.
- Figure 11 (2016 Country-Level, System (All Relationships)) is generated by selecting “f,” “s,” “c,” the country code of the respondent bank, “USD,” and “no” for using an additional index.
  - Visualization notes from Figure 11: Node size reflects PageRank score, edge width reflects normalized flow size; white node color reflects the selected country, and green edges reflect new relationships compared to previous period. Node labels (ISO 2-letter country code) have been turned off for visibility.
- Figure 12 (2016 Country-Level, System (All Relationships) with index) is generated by selecting “f,” “s,” “c,” the country code of the respondent bank, “USD,” and “yes” for using an additional index (saved in a CSV file), in this case the Corruption Perception Index (CPI) published by Transparency International (Annex IV).
  - Visualization mapping for Figure 12:
    - Inside node color: white for domestic country, orange and red for second-lowest and lowest quintiles, respectively (position in the relevant period’s CPI).
    - Outside node color: green if the country node’s CPI score improved, red if it worsened over the previous period.
    - Green edges reflect new relationships compared to previous period.
  - Additional CPI-based encoding described: red (countries in the lowest quintile on the CPI), orange (countries in the 3rd quintile of the CPI), grey (all other countries). Outer ring colored green (improved), black (same ranking), or red (lower ranking) compared to previous year.

### Use of additional index and customization
- color_using_index can be “yes” or “no”; when “yes” an additional CSV file with country-based data (e.g., CPI) is required.
- With an additional index, countries receive colors and rim colors based on index quintile and change from previous period.
- The underlying code can be modified to:
  - incorporate different country-based information (examples listed: Macroeconomic data, various country risk indicators, cyber risk threat data);
  - simulate the impact of various scenarios (similar to a stress test), though scenario simulation is beyond the scope of the paper.

### Implementation notes and templates
- The SWIFT-derived maps are customizable to account for jurisdictional differences in banking system complexity and correspondent banking relationships.
- The Minimum Scope Framework and Expanded Scope Framework are presented elsewhere in the paper; Expanded Scope data can be used to cross-check Minimum Scope Template results but requires additional collection cost and capacity.
- Annex I.A: Minimum Scope Template columns and drop-down lists described (Account description, Restrictions, Type of Restrictions) with enumerated options (1–11 for Account description; 1–5 for Restrictions; 1–5 for Type of Restrictions).
- Annex I.B: Expanded Scope Template column mapping and formatting guidance:
  - The Expanded Scope Template should be reported as a separate .csv by reporting institutions.
  - Fields with numbers should be formatted not to include commas and/or decimals (example formats noted but not altered).
  - Date field must be text formatted (e.g., Mar-17).
- Annex I.C: Clustering of BIC8 codes guidance:
  - To create a Global CBR, replace BIC8 codes with the first four letters of the main BIC8 (example: BOFA for Bank of America entities).
  - Clustering may require knowledge of group structure for banks where first four letters differ due to mergers (examples provided: Wachovia/PNBP and Wells Fargo/WFBI; Deutsche Bank BKTR and DEUT).
  - Clustering does not affect country code or currency analysis because those are in separate columns.

*Source: wp17216 - Box 1. Network Analysis: Selection Menu in Program Code*

### ANNEX II: PROGRAMMING CODE ANALYTICAL TOOL FOR SWIFT DATA ANALYSIS

### ANNEX II: PROGRAMMING CODE ANALYTICAL TOOL FOR SWIFT DATA ANALYSIS

### Network analysis: qualitative and quantitative mapping of financial institutions
- Network analysis represents financial institutions as nodes and flows between them as edges; node size is used to represent a node’s centrality score.
- Centrality indicates critical positions in the network and different centrality measures capture different types of importance:
  - Betweenness: A node with high betweennness lies on a high proportion of paths
  - Closeness: A node with high closeness can transact quickly with nodes in the network.
  - Degree: A node with high degree has many relationships.
  - Eigenvector: A node with a high eigenvector is connected to important neighbors.
- Note: PageRank is a simplified version of Eigenvector centrality, taking less time to compute (Brin; Page, 1998).
- Practical insights from the Krackhardt kite graph example (ten nodes, banks numbered 0–9, each edge reflects the same amount):
  - Bank 3 has the highest degree (most relationships).
  - Failure of Bank 3 would not disconnect payments between other banks.
  - Bank 7 has the highest betweenness; its failure would break flows between Banks 8 (and 9) and the rest of the network.
  - Banks 5 and 6 have the highest closeness; flows from them take the fewest steps to reach all points in the network.
- Implementation note: While the Extended Scope Framework uses PageRank as a metric for the relative size of nodes, any centrality measure could be used instead depending on which information about the banks is most useful.

### Country-level overlay data: CPI list used for country data analysis
- Purpose: Add country-based information (for example, Corruption Perceptions Index (CPI) scores) as layers onto country network maps to display multiple pieces of information simultaneously (e.g., node interior color from yearly CPI scores and rim color from yearly changes).
- Data structure described:
  - A .csv file containing the ranking of 176 countries on the Corruption Perceptions Index (CPI) published by Transparency International.
  - First column: ISO two-letter country code (to be merged with SWIFT country data).
  - Columns s2012 – s2016: yearly ranking of the countries (0 when no score is available).
  - Columns ch2013 – ch2016: change in the score compared with the previous year (0 if same, 1 if score increased, -1 if the score decreased, and empty cell if no score was available).
- Sample CPI table header and rows (as presented in the source):
  - Codes2012s2013s2014s2015s2016ch2012ch2013ch2014ch2015ch2016
  - AF88121115001-11
  - AL33313336390-1111
  - DZ34363636340100-1
  - AO222319151801-1-11
  - AR35343432360-10-11
  - AM3436373533011-1-1
  - AU85818079790-1-1-10
  - AT69697276750011-1
  - AZ272829293001101
  - BS717171066000
  - BH51484951430-111-1
  - BD262725252601-101
  - BB7675740610-1-1
  - BY31293132400-1111
  - BE757576777700110
  - BJ3636393736001-1-1
  - BT636365656500100
  - BO3434353433001-1-1
  - BA424239383900-1-11
  - BW65646363600-1-10-1
  - BR43424338400-11-11
  - BN5560005801
  - BG4141434141001-10
  - BF383838384200001
  - BI192120212001-11-1
  - KH22202121210-1100
  - CM26252727260-110-1
  - CA84818183820-101-1
  - CV60585755590-1-1-11
  - CF26252424200-1-10-1
  - TD19192222200010-1
  - CL72717370660-11-1-1
  - CN394036374001-111
  - CO363637373700100
  - KM282826262400-10-1
  - CG26222323200-110-1
  - CR54535455580-1111
  - CI29273232340-1101
  - HR46484851490101-1
  - CU48464647470-1010
  - CY66636361550-10-1-1
  - CZ49485156550-111-1
  - CD21222222210100-1
  - DK9091929190011-1-1
  - DJ363634343000-10-1
  - DM585858059000

*Source: ANNEX II: PROGRAMMING CODE ANALYTICAL TOOL FOR SWIFT DATA ANALYSIS*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17216.pdf_